fix(tools): preserve canonical output boundaries

docs/config-catalog.md

@@ -1303,7 +1303,7 @@ export interface Config {
 export type ToolPresentationMode = 'native' | 'code' | 'both'
 ```
 
-Source: [`packages/core/tools/src/index.ts:448`](../packages/core/tools/src/index.ts)
+Source: [`packages/core/tools/src/index.ts:467`](../packages/core/tools/src/index.ts)
 
 ## `@deepseek-ai/dsh-tui`
 

docs/cordis-catalog/services.md

@@ -1361,7 +1361,7 @@ async execute(exec: ToolExecutionInput): Promise<ToolExecutionResult>
 
 Types: [ScopeKey](../core-data-structures/scope.md) · [ToolDefinition](../core-data-structures/tools.md) · [ToolExecutionInput](../core-data-structures/tools.md) · [ToolExecutionMode](../core-data-structures/tools.md) · [ToolExecutionResult](../core-data-structures/tools.md) · [ToolGuard](../core-data-structures/tools.md) · [ToolRestriction](../core-data-structures/tools.md) · [ToolSchema](../core-data-structures/tools.md)
 
-Source: [`packages/core/tools/src/index.ts:504`](../../packages/core/tools/src/index.ts)
+Source: [`packages/core/tools/src/index.ts:523`](../../packages/core/tools/src/index.ts)
 
 ## `ctx.userInteraction` — `UserInteractionService`
 

docs/core-data-structures/session.md

@@ -73,12 +73,13 @@ interface SessionEventMap {
    */
   'tool/call': { turn: number; step: number; callId: CallId; name: string; arguments: string }
   /**
-   * A completed tool call's model-facing result, canonical failure detail, and
-   * optional tool-private `meta` presentation payload. `meta` is opaque to the
-   * core (the producing tool owns its shape and reads it back in `presentResult`)
-   * but MUST be JSON-serializable: `Session.append` runtime-validates all event
-   * data with `isJsonValue`, so a non-serializable `meta` is rejected at the
-   * source, and the durable log reproduces the identical card on replay. Absent
+   * A completed tool call's model-facing result, optional internal failure
+   * identity, and optional tool-private `meta` presentation payload. `meta` is
+   * opaque to the core (the producing tool owns its shape and reads it back in
+   * `presentResult`) but MUST be JSON-serializable: `Session.append`
+   * runtime-validates all event data with `isJsonValue`, so a non-serializable
+   * `meta` is rejected at the source, and the durable log reproduces the
+   * identical card on replay. Absent
    * unless the tool attaches one (e.g. `dsh-tool-fs` carries its result-time
    * contextual diff here).
    */
@@ -88,7 +89,7 @@ interface SessionEventMap {
     callId: CallId
     content: ContentBlock[]
     isError: boolean
-    error?: { message: string; info?: { name: string; code: string } }
+    error?: { name: string; code: string }
     meta?: JsonValue
   }
   /** Steering content injected between steps of a running turn. */

docs/persistence-catalog.md

@@ -79,7 +79,7 @@ export type SessionEvent<T extends SessionEventType = SessionEventType> = {
 }[T]
 ```
 
-Sources: [`packages/core/session/src/types.ts:276`](../packages/core/session/src/types.ts) · [`packages/core/session/src/types.ts:283`](../packages/core/session/src/types.ts) · [`packages/core/session/src/types.ts:313`](../packages/core/session/src/types.ts) · [`packages/core/session/src/types.ts:345`](../packages/core/session/src/types.ts)
+Sources: [`packages/core/session/src/types.ts:277`](../packages/core/session/src/types.ts) · [`packages/core/session/src/types.ts:284`](../packages/core/session/src/types.ts) · [`packages/core/session/src/types.ts:314`](../packages/core/session/src/types.ts) · [`packages/core/session/src/types.ts:346`](../packages/core/session/src/types.ts)
 
 ## Events
 
@@ -356,7 +356,7 @@ Source: [`packages/core/session/src/types.ts:213`](../packages/core/session/src/
 'request/header': { header: EpochHeader; reason: RequestHeaderReason }
 ```
 
-Source: [`packages/core/session/src/types.ts:272`](../packages/core/session/src/types.ts)
+Source: [`packages/core/session/src/types.ts:273`](../packages/core/session/src/types.ts)
 
 ### `sandbox/*`
 
@@ -387,7 +387,7 @@ Source: [`packages/sandbox/sandbox-policy/src/session-mode.ts:34`](../packages/s
 
 Types: [ContentBlock](core-data-structures/core.md) · [MessageSource](core-data-structures/core.md)
 
-Source: [`packages/core/session/src/types.ts:265`](../packages/core/session/src/types.ts)
+Source: [`packages/core/session/src/types.ts:266`](../packages/core/session/src/types.ts)
 
 ### `step/*`
 
@@ -420,7 +420,7 @@ Source: [`packages/core/session/src/types.ts:204`](../packages/core/session/src/
 
 Types: [TodoItem](core-data-structures/session.md)
 
-Source: [`packages/core/session/src/types.ts:267`](../packages/core/session/src/types.ts)
+Source: [`packages/core/session/src/types.ts:268`](../packages/core/session/src/types.ts)
 
 ### `tool/*`
 
@@ -468,12 +468,13 @@ Source: [`packages/core/tools/src/code-mode.ts:34`](../packages/core/tools/src/c
 
 ```ts persistence-catalog
 /**
- * A completed tool call's model-facing result, canonical failure detail, and
- * optional tool-private `meta` presentation payload. `meta` is opaque to the
- * core (the producing tool owns its shape and reads it back in `presentResult`)
- * but MUST be JSON-serializable: `Session.append` runtime-validates all event
- * data with `isJsonValue`, so a non-serializable `meta` is rejected at the
- * source, and the durable log reproduces the identical card on replay. Absent
+ * A completed tool call's model-facing result, optional internal failure
+ * identity, and optional tool-private `meta` presentation payload. `meta` is
+ * opaque to the core (the producing tool owns its shape and reads it back in
+ * `presentResult`) but MUST be JSON-serializable: `Session.append`
+ * runtime-validates all event data with `isJsonValue`, so a non-serializable
+ * `meta` is rejected at the source, and the durable log reproduces the
+ * identical card on replay. Absent
  * unless the tool attaches one (e.g. `dsh-tool-fs` carries its result-time
  * contextual diff here).
  */
@@ -483,14 +484,14 @@ Source: [`packages/core/tools/src/code-mode.ts:34`](../packages/core/tools/src/c
   callId: CallId
   content: ContentBlock[]
   isError: boolean
-  error?: { message: string; info?: { name: string; code: string } }
+  error?: { name: string; code: string }
   meta?: JsonValue
 }
 ```
 
 Types: [CallId](core-data-structures/core.md) · [ContentBlock](core-data-structures/core.md)
 
-Source: [`packages/core/session/src/types.ts:255`](../packages/core/session/src/types.ts)
+Source: [`packages/core/session/src/types.ts:256`](../packages/core/session/src/types.ts)
 
 ### `turn/*`
 

examples/acp-agent/tests/snapshots/cancel-tool-calls/session.jsonl

@@ -13,8 +13,8 @@
 {"type":"assistant/chunk","seq":11,"time":1784437195077,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}}
 {"type":"assistant/message","seq":12,"time":1784437195078,"data":{"turn":1,"step":1,"content":[{"type":"tool-call","id":"call_wait","name":"bash","arguments":"{\"command\":\"node -e \\\"setInterval(() => {}, 1000)\\\"\",\"description\":\"Wait until cancellation\"}"},{"type":"tool-call","id":"call_skipped","name":"bash","arguments":"{\"command\":\"printf skipped > skipped.txt\",\"description\":\"Write skipped marker\"}"}],"provenance":{"provider":"deepseek","model":"deepseek-v4-flash"},"usage":{"inputTokens":10,"outputTokens":10}},"sourceEventSeqs":[4,5,6,7,8,9,10,11],"surfaceOp":"append"}
 {"type":"tool/call","seq":13,"time":1784437195078,"data":{"turn":1,"step":1,"callId":"call_wait","name":"bash","arguments":"{\"command\":\"node -e \\\"setInterval(() => {}, 1000)\\\"\",\"description\":\"Wait until cancellation\"}"}}
-{"type":"tool/result","seq":14,"time":1784437195089,"data":{"turn":1,"step":1,"callId":"call_wait","content":[{"type":"text","text":"Error: command aborted"}],"isError":true,"error":{"message":"command aborted"}},"sourceEventSeqs":[13],"surfaceOp":"append"}
+{"type":"tool/result","seq":14,"time":1784437195089,"data":{"turn":1,"step":1,"callId":"call_wait","content":[{"type":"text","text":"Error: command aborted"}],"isError":true},"sourceEventSeqs":[13],"surfaceOp":"append"}
 {"type":"tool/call","seq":15,"time":1784437195089,"data":{"turn":1,"step":1,"callId":"call_skipped","name":"bash","arguments":"{\"command\":\"printf skipped > skipped.txt\",\"description\":\"Write skipped marker\"}"}}
-{"type":"tool/result","seq":16,"time":1784437195089,"data":{"turn":1,"step":1,"callId":"call_skipped","content":[{"type":"text","text":"Error: tool call skipped because the step was aborted before execution"}],"isError":true,"error":{"message":"tool call skipped because the step was aborted before execution","info":{"name":"AbortError","code":"ABORTED"}}},"sourceEventSeqs":[15],"surfaceOp":"append"}
+{"type":"tool/result","seq":16,"time":1784437195089,"data":{"turn":1,"step":1,"callId":"call_skipped","content":[{"type":"text","text":"Error: tool call skipped because the step was aborted before execution"}],"isError":true,"error":{"name":"AbortError","code":"ABORTED"}},"sourceEventSeqs":[15],"surfaceOp":"append"}
 {"type":"step/end","seq":17,"time":1784437195090,"data":{"turn":1,"step":1}}
 {"type":"turn/end","seq":18,"time":1784437195090,"data":{"turn":1,"reason":{"kind":"aborted","reason":"session/cancel"}}}

examples/acp-agent/tests/snapshots/escalation-rejected/session.jsonl

@@ -157,7 +157,7 @@
 {"type":"tool/call","seq":155,"time":1783962246274,"data":{"turn":1,"step":1,"callId":"call_00_WB1vnPomi8yr6MlcFKTj7912","name":"bash","arguments":"{\"command\": \"printf 'escalated\\\\n' > /tmp/dsh-escalated.txt && cat /tmp/dsh-escalated.txt && rm /tmp/dsh-escalated.txt\", \"description\": \"Write to /tmp and verify, then clean up\", \"sandbox_permissions\": \"danger-full-access\", \"justification\": \"the user asked to write a file outside the workspace\"}"}}
 {"type":"approval/asked","seq":156,"time":1783962246275,"data":{"id":"46c8dba4-52c9-4a6a-b6a6-5f34c95c28df","toolName":"bash","callId":"call_00_WB1vnPomi8yr6MlcFKTj7912","reason":"escalate sandbox to danger-full-access: the user asked to write a file outside the workspace"}}
 {"type":"approval/decided","seq":157,"time":1783962246275,"data":{"id":"46c8dba4-52c9-4a6a-b6a6-5f34c95c28df","outcome":"rejected"}}
-{"type":"tool/result","seq":158,"time":1783962246275,"data":{"turn":1,"step":1,"callId":"call_00_WB1vnPomi8yr6MlcFKTj7912","content":[{"type":"text","text":"Error: the user rejected escalating this command to \"danger-full-access\""}],"isError":true,"error":{"message":"the user rejected escalating this command to \"danger-full-access\""}},"sourceEventSeqs":[155],"surfaceOp":"append"}
+{"type":"tool/result","seq":158,"time":1783962246275,"data":{"turn":1,"step":1,"callId":"call_00_WB1vnPomi8yr6MlcFKTj7912","content":[{"type":"text","text":"Error: the user rejected escalating this command to \"danger-full-access\""}],"isError":true},"sourceEventSeqs":[155],"surfaceOp":"append"}
 {"type":"step/end","seq":159,"time":1783962246276,"data":{"turn":1,"step":1}}
 {"type":"step/start","seq":160,"time":1783962246276,"data":{"turn":1,"step":2}}
 {"type":"assistant/chunk","seq":161,"time":1783860683140,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}}

examples/acp-agent/tests/snapshots/fs-policy-reject/session.jsonl

@@ -77,7 +77,7 @@
 {"type":"assistant/chunk","seq":75,"time":1783611703969,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}}
 {"type":"assistant/message","seq":76,"time":1783611703972,"data":{"turn":1,"step":1,"content":[{"type":"reasoning","text":"The user wants me to use the edit tool to replace \"blue\" with \"green\" in settings.txt without reading the file first, and then reply with just \"DONE\"."},{"type":"tool-call","id":"call_00_x0zlnXl5JOxLrAYL9y7P0119","name":"edit","arguments":"{\"file_path\": \"settings.txt\", \"old_string\": \"blue\", \"new_string\": \"green\"}"}],"provenance":{"provider":"deepseek","model":"deepseek-v4-flash"},"usage":{"inputTokens":3132,"outputTokens":115,"cacheReadTokens":0,"reasoningTokens":36}},"sourceEventSeqs":[4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75],"surfaceOp":"append"}
 {"type":"tool/call","seq":77,"time":1783611703972,"data":{"turn":1,"step":1,"callId":"call_00_x0zlnXl5JOxLrAYL9y7P0119","name":"edit","arguments":"{\"file_path\": \"settings.txt\", \"old_string\": \"blue\", \"new_string\": \"green\"}"}}
-{"type":"tool/result","seq":78,"time":1783611703978,"data":{"turn":1,"step":1,"callId":"call_00_x0zlnXl5JOxLrAYL9y7P0119","content":[{"type":"text","text":"Error: edit requires reading \"/var/folders/2c/psb0_fmx7hbgz558xjt_f0l00000gn/T/acp-snap-cwd-QzoqnB/settings.txt\" first"}],"isError":true,"error":{"message":"edit requires reading \"/var/folders/2c/psb0_fmx7hbgz558xjt_f0l00000gn/T/acp-snap-cwd-QzoqnB/settings.txt\" first","info":{"name":"FsError","code":"FS_NOT_OBSERVED"}}},"sourceEventSeqs":[77],"surfaceOp":"append"}
+{"type":"tool/result","seq":78,"time":1783611703978,"data":{"turn":1,"step":1,"callId":"call_00_x0zlnXl5JOxLrAYL9y7P0119","content":[{"type":"text","text":"Error: edit requires reading \"/var/folders/2c/psb0_fmx7hbgz558xjt_f0l00000gn/T/acp-snap-cwd-QzoqnB/settings.txt\" first"}],"isError":true,"error":{"name":"FsError","code":"FS_NOT_OBSERVED"}},"sourceEventSeqs":[77],"surfaceOp":"append"}
 {"type":"step/end","seq":79,"time":1783611703978,"data":{"turn":1,"step":1}}
 {"type":"step/start","seq":80,"time":1783611703978,"data":{"turn":1,"step":2}}
 {"type":"assistant/chunk","seq":81,"time":1783611704825,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}}

examples/acp-agent/tests/snapshots/hook-cc-posttool-block/session.jsonl

@@ -75,7 +75,7 @@
 {"type":"tool/call","seq":73,"time":1783962505993,"data":{"turn":1,"step":1,"callId":"call_00_VAByyMjsct4c7P6k1ysX9256","name":"bash","arguments":"{\"command\": \"echo HELLO\", \"description\": \"Echo HELLO to stdout\"}"}}
 {"type":"hook/invoked","seq":74,"time":1783962506001,"data":{"turn":1,"point":"PostToolUse","dialect":"claude","handlerId":"claude:PostToolUse:1","matcher":"bash"}}
 {"type":"hook/result","seq":75,"time":1783962506011,"data":{"turn":1,"point":"PostToolUse","handlerId":"claude:PostToolUse:1","decision":"block","exitCode":2,"stderrSummary":"tool output rejected by policy: retry once","durationMs":9.922291999999743}}
-{"type":"tool/result","seq":76,"time":1783962506011,"data":{"turn":1,"step":1,"callId":"call_00_VAByyMjsct4c7P6k1ysX9256","content":[{"type":"text","text":"tool output rejected by policy: retry once"}],"isError":true,"error":{"message":"tool output rejected by policy: retry once"}},"sourceEventSeqs":[73],"surfaceOp":"append"}
+{"type":"tool/result","seq":76,"time":1783962506011,"data":{"turn":1,"step":1,"callId":"call_00_VAByyMjsct4c7P6k1ysX9256","content":[{"type":"text","text":"tool output rejected by policy: retry once"}],"isError":true},"sourceEventSeqs":[73],"surfaceOp":"append"}
 {"type":"step/end","seq":77,"time":1783962506012,"data":{"turn":1,"step":1}}
 {"type":"step/start","seq":78,"time":1783962506012,"data":{"turn":1,"step":2}}
 {"type":"assistant/chunk","seq":79,"time":1783962507038,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}}

examples/acp-agent/tests/snapshots/hook-cc-pretool-ask/session.jsonl

@@ -57,7 +57,7 @@
 {"type":"hook/result","seq":55,"time":1783352172573,"data":{"turn":1,"point":"PreToolUse","handlerId":"claude:PreToolUse:1","decision":"ask","exitCode":0,"durationMs":14.113374999999905}}
 {"type":"approval/asked","seq":56,"time":1783962235813,"data":{"id":"68f1e09d-f3e5-4e39-8a51-da082ba3ba99","toolName":"bash","callId":"call_00_6k0oGSliVHxGSgqBmMEO4311","reason":"bash requires manual approval in this session"}}
 {"type":"approval/decided","seq":57,"time":1783962235813,"data":{"id":"68f1e09d-f3e5-4e39-8a51-da082ba3ba99","outcome":"rejected"}}
-{"type":"tool/result","seq":58,"time":1783962235814,"data":{"turn":1,"step":1,"callId":"call_00_6k0oGSliVHxGSgqBmMEO4311","content":[{"type":"text","text":"Error: the user rejected tool \"bash\""}],"isError":true,"error":{"message":"the user rejected tool \"bash\""}},"sourceEventSeqs":[53],"surfaceOp":"append"}
+{"type":"tool/result","seq":58,"time":1783962235814,"data":{"turn":1,"step":1,"callId":"call_00_6k0oGSliVHxGSgqBmMEO4311","content":[{"type":"text","text":"Error: the user rejected tool \"bash\""}],"isError":true},"sourceEventSeqs":[53],"surfaceOp":"append"}
 {"type":"step/end","seq":59,"time":1783962235814,"data":{"turn":1,"step":1}}
 {"type":"step/start","seq":60,"time":1783962235814,"data":{"turn":1,"step":2}}
 {"type":"assistant/chunk","seq":61,"time":1783352173584,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}}

examples/acp-agent/tests/snapshots/hook-cc-pretool-deny/session.jsonl

@@ -55,7 +55,7 @@
 {"type":"tool/call","seq":53,"time":1783352166514,"data":{"turn":1,"step":1,"callId":"call_00_JliP571Bh0QQ8QExbSPk0080","name":"bash","arguments":"{\"command\": \"echo HELLO\", \"description\": \"Run echo HELLO\"}"}}
 {"type":"hook/invoked","seq":54,"time":1783352166515,"data":{"turn":1,"point":"PreToolUse","dialect":"claude","handlerId":"claude:PreToolUse:1","matcher":"bash"}}
 {"type":"hook/result","seq":55,"time":1783352166528,"data":{"turn":1,"point":"PreToolUse","handlerId":"claude:PreToolUse:1","decision":"block","exitCode":2,"stderrSummary":"bash is disabled by policy in this session","durationMs":12.367074000000684}}
-{"type":"tool/result","seq":56,"time":1783352166528,"data":{"turn":1,"step":1,"callId":"call_00_JliP571Bh0QQ8QExbSPk0080","content":[{"type":"text","text":"Error: bash is disabled by policy in this session"}],"isError":true,"error":{"message":"bash is disabled by policy in this session"}},"sourceEventSeqs":[53],"surfaceOp":"append"}
+{"type":"tool/result","seq":56,"time":1783352166528,"data":{"turn":1,"step":1,"callId":"call_00_JliP571Bh0QQ8QExbSPk0080","content":[{"type":"text","text":"Error: bash is disabled by policy in this session"}],"isError":true},"sourceEventSeqs":[53],"surfaceOp":"append"}
 {"type":"step/end","seq":57,"time":1783352166529,"data":{"turn":1,"step":1}}
 {"type":"step/start","seq":58,"time":1783352166529,"data":{"turn":1,"step":2}}
 {"type":"assistant/chunk","seq":59,"time":1783352167307,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}}

examples/acp-agent/tests/snapshots/hook-codex-posttool-block/session.jsonl

@@ -66,7 +66,7 @@
 {"type":"tool/call","seq":64,"time":1783986963664,"data":{"turn":1,"step":1,"callId":"call_00_1rmSWHhVchVg7PDTmegT0421","name":"bash","arguments":"{\"command\": \"echo HELLO\", \"description\": \"Echo HELLO to stdout\"}"}}
 {"type":"hook/invoked","seq":65,"time":1783986963673,"data":{"turn":1,"point":"PostToolUse","dialect":"codex","handlerId":"codex:PostToolUse:1","matcher":"bash"}}
 {"type":"hook/result","seq":66,"time":1783986963677,"data":{"turn":1,"point":"PostToolUse","handlerId":"codex:PostToolUse:1","decision":"block","exitCode":2,"stderrSummary":"tool output rejected by codex policy: summarize instead","durationMs":4.42941699999983}}
-{"type":"tool/result","seq":67,"time":1783986963678,"data":{"turn":1,"step":1,"callId":"call_00_1rmSWHhVchVg7PDTmegT0421","content":[{"type":"text","text":"tool output rejected by codex policy: summarize instead"}],"isError":true,"error":{"message":"tool output rejected by codex policy: summarize instead"}},"sourceEventSeqs":[64],"surfaceOp":"append"}
+{"type":"tool/result","seq":67,"time":1783986963678,"data":{"turn":1,"step":1,"callId":"call_00_1rmSWHhVchVg7PDTmegT0421","content":[{"type":"text","text":"tool output rejected by codex policy: summarize instead"}],"isError":true},"sourceEventSeqs":[64],"surfaceOp":"append"}
 {"type":"step/end","seq":68,"time":1783986963678,"data":{"turn":1,"step":1}}
 {"type":"step/start","seq":69,"time":1783986963679,"data":{"turn":1,"step":2}}
 {"type":"assistant/chunk","seq":70,"time":1783986964555,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}}

examples/acp-agent/tests/snapshots/hook-codex-pretool-block/session.jsonl

@@ -55,7 +55,7 @@
 {"type":"tool/call","seq":53,"time":1783352215804,"data":{"turn":1,"step":1,"callId":"call_00_tv0SMeLXaTuyuVrOxnV97085","name":"bash","arguments":"{\"command\": \"echo HELLO\", \"description\": \"Run echo HELLO\"}"}}
 {"type":"hook/invoked","seq":54,"time":1783352215805,"data":{"turn":1,"point":"PreToolUse","dialect":"codex","handlerId":"codex:PreToolUse:1","matcher":"bash"}}
 {"type":"hook/result","seq":55,"time":1783352215832,"data":{"turn":1,"point":"PreToolUse","handlerId":"codex:PreToolUse:1","decision":"block","exitCode":2,"stderrSummary":"bash is disabled by codex policy in this session","durationMs":26.08518500000082}}
-{"type":"tool/result","seq":56,"time":1783352215832,"data":{"turn":1,"step":1,"callId":"call_00_tv0SMeLXaTuyuVrOxnV97085","content":[{"type":"text","text":"Error: bash is disabled by codex policy in this session"}],"isError":true,"error":{"message":"bash is disabled by codex policy in this session"}},"sourceEventSeqs":[53],"surfaceOp":"append"}
+{"type":"tool/result","seq":56,"time":1783352215832,"data":{"turn":1,"step":1,"callId":"call_00_tv0SMeLXaTuyuVrOxnV97085","content":[{"type":"text","text":"Error: bash is disabled by codex policy in this session"}],"isError":true},"sourceEventSeqs":[53],"surfaceOp":"append"}
 {"type":"step/end","seq":57,"time":1783352215833,"data":{"turn":1,"step":1}}
 {"type":"step/start","seq":58,"time":1783352215834,"data":{"turn":1,"step":2}}
 {"type":"assistant/chunk","seq":59,"time":1783352216779,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"reasoning"}}}

examples/acp-agent/tests/snapshots/subagent-depth-two-rejection/session.2.jsonl

@@ -10,7 +10,7 @@
 {"type":"assistant/chunk","seq":8,"time":0,"data":{"turn":1,"step":1,"chunk":{"type":"finish","reason":{"kind":"tool-calls"}}}}
 {"type":"assistant/message","seq":9,"time":1784540790335,"data":{"turn":1,"step":1,"content":[{"type":"tool-call","id":"call_depth_three_rejected","name":"subagent","arguments":"{\"description\":\"Exceed depth cap\",\"prompt\":\"This child must never start.\"}"}],"provenance":{"provider":"deepseek","model":"deepseek-v4-flash"},"usage":{"inputTokens":10,"outputTokens":5}},"sourceEventSeqs":[4,5,6,7,8],"surfaceOp":"append"}
 {"type":"tool/call","seq":10,"time":1784540790335,"data":{"turn":1,"step":1,"callId":"call_depth_three_rejected","name":"subagent","arguments":"{\"description\":\"Exceed depth cap\",\"prompt\":\"This child must never start.\"}"}}
-{"type":"tool/result","seq":11,"time":1784540790337,"data":{"turn":1,"step":1,"callId":"call_depth_three_rejected","content":[{"type":"text","text":"Error: subagent depth 3 exceeds maxDepth 2"}],"isError":true,"error":{"message":"subagent depth 3 exceeds maxDepth 2"}},"sourceEventSeqs":[10],"surfaceOp":"append"}
+{"type":"tool/result","seq":11,"time":1784540790337,"data":{"turn":1,"step":1,"callId":"call_depth_three_rejected","content":[{"type":"text","text":"Error: subagent depth 3 exceeds maxDepth 2"}],"isError":true},"sourceEventSeqs":[10],"surfaceOp":"append"}
 {"type":"step/end","seq":12,"time":1784540790338,"data":{"turn":1,"step":1}}
 {"type":"step/start","seq":13,"time":1784540790338,"data":{"turn":1,"step":2}}
 {"type":"assistant/chunk","seq":14,"time":1784540790339,"data":{"turn":1,"step":2,"chunk":{"type":"block-start","index":0,"blockType":"text"}}}

packages/compact/compact-tool-result-prune/tests/tool-result-prune.spec.ts

@@ -151,7 +151,7 @@ describe('ToolResultPruneService session transaction', () => {
       text: 'x'.repeat(100),
     }], {
       isError: true,
-      error: { message: 'exit 1', info: { name: 'ExitError', code: 'EXIT_1' } },
+      error: { name: 'ExitError', code: 'EXIT_1' },
       meta: { diff: ['a', 'b'] },
       futureField: { nested: true },
     })
@@ -180,7 +180,7 @@ describe('ToolResultPruneService session transaction', () => {
         step: 1,
         callId: CallId('one'),
         isError: true,
-        error: { message: 'exit 1', info: { name: 'ExitError', code: 'EXIT_1' } },
+        error: { name: 'ExitError', code: 'EXIT_1' },
         meta: { diff: ['a', 'b'] },
         futureField: { nested: true },
       },

packages/cordis/tool-cordis/src/api-catalog.ts

@@ -1450,7 +1450,7 @@ export const TYPE_API: readonly TypeApiEntry[] = [
   },
   {
     name: 'SessionEventMap',
-    declaration: 'export interface SessionEventMap {\n    \'turn/start\': {\n        turn: number;\n        trigger: TurnTrigger;\n    };\n    \'turn/end\': {\n        turn: number;\n        reason: TurnEndReason;\n    };\n    \'step/start\': {\n        turn: number;\n        step: number;\n    };\n    \'step/end\': {\n        turn: number;\n        step: number;\n    };\n    \'user/message\': {\n        content: ContentBlock[];\n        source: MessageSource;\n    };\n    \'prompt/blocked\': {\n        content: ContentBlock[];\n        source: MessageSource;\n        reason: string;\n    };\n    \'context/message\': {\n        content: ContentBlock[];\n        source: MessageSource;\n        meta?: JsonValue;\n    };\n    \'assistant/chunk\': {\n        turn: number;\n        step: number;\n        chunk: StreamChunk;\n    };\n    \'assistant/message\': {\n        turn: number;\n        step: number;\n        content: ContentBlock[];\n        provenance: AssistantProvenance;\n        usage?: TokenUsage;\n    };\n    \'tool/call\': {\n        turn: number;\n        step: number;\n        callId: CallId;\n        name: string;\n        arguments: string;\n    };\n    \'tool/result\': {\n        turn: number;\n        step: number;\n        callId: CallId;\n        content: ContentBlock[];\n        isError: boolean;\n        error?: {\n            message: string;\n            info?: {\n                name: string;\n                code: string;\n            };\n        };\n        meta?: JsonValue;\n    };\n    \'steering/message\': {\n        turn: number;\n        content: Content /* …truncated — full shape in source */',
+    declaration: 'export interface SessionEventMap {\n    \'turn/start\': {\n        turn: number;\n        trigger: TurnTrigger;\n    };\n    \'turn/end\': {\n        turn: number;\n        reason: TurnEndReason;\n    };\n    \'step/start\': {\n        turn: number;\n        step: number;\n    };\n    \'step/end\': {\n        turn: number;\n        step: number;\n    };\n    \'user/message\': {\n        content: ContentBlock[];\n        source: MessageSource;\n    };\n    \'prompt/blocked\': {\n        content: ContentBlock[];\n        source: MessageSource;\n        reason: string;\n    };\n    \'context/message\': {\n        content: ContentBlock[];\n        source: MessageSource;\n        meta?: JsonValue;\n    };\n    \'assistant/chunk\': {\n        turn: number;\n        step: number;\n        chunk: StreamChunk;\n    };\n    \'assistant/message\': {\n        turn: number;\n        step: number;\n        content: ContentBlock[];\n        provenance: AssistantProvenance;\n        usage?: TokenUsage;\n    };\n    \'tool/call\': {\n        turn: number;\n        step: number;\n        callId: CallId;\n        name: string;\n        arguments: string;\n    };\n    \'tool/result\': {\n        turn: number;\n        step: number;\n        callId: CallId;\n        content: ContentBlock[];\n        isError: boolean;\n        error?: {\n            name: string;\n            code: string;\n        };\n        meta?: JsonValue;\n    };\n    \'steering/message\': {\n        turn: number;\n        content: ContentBlock[];\n        source: MessageSource;\n    };\n    \'todo/write\': {\n       /* …truncated — full shape in source */',
   },
   {
     name: 'SessionEventReadRequest',

packages/core/agent-loop/src/tool-calls.ts

@@ -248,7 +248,7 @@ function appendToolResult(
     callId: block.id,
     content: result.content,
     isError: result.isError,
-    ...result.error ? { error: result.error } : {},
+    ...result.error?.info ? { error: result.error.info } : {},
     // The tool's private presentation payload (e.g. a result-time diff),
     // persisted so a UI bridge reproduces the card on replay.
     ...result.meta !== undefined ? { meta: result.meta } : {},

packages/core/agent-loop/tests/cancel.spec.ts

@@ -397,7 +397,7 @@ describe('Agent.cancel()', () => {
     expect(result?.type === 'tool/result' ? result.data : undefined).toMatchObject({
       callId: 'c1',
       isError: true,
-      error: { info: { name: 'AbortError', code: 'ABORTED' } },
+      error: { name: 'AbortError', code: 'ABORTED' },
     })
 
     send(agent, 'continue safely')

packages/core/agent-loop/tests/contract-regressions.spec.ts

@@ -259,7 +259,7 @@ describe('abort during tool execution ends the turn', () => {
         case 'assistant/message': order.push('assistant/message'); break
         case 'tool/call': order.push(`tool/call:${event.data.callId}`); break
         case 'tool/result': {
-          const outcome = event.data.error?.info?.code === 'ABORTED' ? 'synthetic-aborted' : 'real'
+          const outcome = event.data.error?.code === 'ABORTED' ? 'synthetic-aborted' : 'real'
           order.push(`tool/result:${event.data.callId}:${outcome}`)
           break
         }
@@ -308,7 +308,7 @@ describe('abort during tool execution ends the turn', () => {
     expect(results[1]!.data).toMatchObject({
       callId: CallId('c2'),
       isError: true,
-      error: { info: { name: 'AbortError', code: 'ABORTED' } },
+      error: { name: 'AbortError', code: 'ABORTED' },
     })
   })
 

packages/core/agent-loop/tests/tool-calls.spec.ts

@@ -479,18 +479,12 @@ describe('tool-call scheduler: abort handling', () => {
       {
         callId: CallId('c1'),
         isError: true,
-        error: {
-          message: 'tool call skipped because the step was aborted before execution',
-          info: { name: 'AbortError', code: 'ABORTED' },
-        },
+        error: { name: 'AbortError', code: 'ABORTED' },
       },
       {
         callId: CallId('c2'),
         isError: true,
-        error: {
-          message: 'tool call skipped because the step was aborted before execution',
-          info: { name: 'AbortError', code: 'ABORTED' },
-        },
+        error: { name: 'AbortError', code: 'ABORTED' },
       },
     ])
   })
@@ -523,7 +517,7 @@ describe('tool-call scheduler: abort handling', () => {
     expect(events(agent).filter(e => e.type === 'tool/result').map(e => e.data.callId))
       .toEqual([CallId('c1'), CallId('c2')])
     expect(events(agent).filter(e => e.type === 'tool/result').at(-1)?.data)
-      .toMatchObject({ callId: CallId('c2'), isError: true, error: { info: { name: 'AbortError', code: 'ABORTED' } } })
+      .toMatchObject({ callId: CallId('c2'), isError: true, error: { name: 'AbortError', code: 'ABORTED' } })
   })
 
   it('stops replenishing after abort, commits started results, and drains accepted additional contexts', async () => {
@@ -555,7 +549,7 @@ describe('tool-call scheduler: abort handling', () => {
     expect(events(agent).filter(e => e.type === 'tool/result').slice(-2).map(e => ({
       callId: e.data.callId,
       isError: e.data.isError,
-      errorInfo: e.data.error?.info,
+      errorInfo: e.data.error,
     })))
       .toEqual([
         { callId: CallId('c3'), isError: true, errorInfo: { name: 'AbortError', code: 'ABORTED' } },
@@ -601,6 +595,6 @@ describe('tool-call scheduler: abort handling', () => {
     expect(events(agent).filter(e => e.type === 'tool/call').map(e => e.data.callId))
       .toEqual([CallId('c1'), CallId('c2'), CallId('c3')])
     expect(events(agent).filter(e => e.type === 'tool/result').at(-1)?.data)
-      .toMatchObject({ callId: CallId('c3'), isError: true, error: { info: { name: 'AbortError', code: 'ABORTED' } } })
+      .toMatchObject({ callId: CallId('c3'), isError: true, error: { name: 'AbortError', code: 'ABORTED' } })
   })
 })

packages/core/session/README.md

@@ -58,7 +58,7 @@ Durable values need one accepted representation, not a check followed by a secon
 
 `context/message` renders its `content` verbatim as a user-role message, and may attach JSON `meta` for replayable plugin state; metadata remains durable but is excluded from `deriveMessages()`.
 
-`tool/result` persists the model-facing content, canonical failure detail, and optional presentation metadata. A tool's successful canonical `value` is deliberately execution-local and never enters the session event, so replay reconstructs the Native/model presentation but cannot recover intermediate programmatic values. This does not change `SESSION_FORMAT_VERSION`: the persisted projection remains authoritative.
+`tool/result` persists the model-facing content, optional internal failure identity, and optional presentation metadata. A tool's successful canonical `value` and human-readable canonical failure message remain execution-local; rendered error content is the replay-authoritative message. This preserves the existing event shape and does not change `SESSION_FORMAT_VERSION`.
 
 ### Session event vocabulary (`types.ts`)
 

packages/core/session/src/repair.ts

@@ -92,10 +92,7 @@ export function interruptedTurnClosers(events: readonly SessionEvent[]): Session
         callId,
         content: [{ type: 'text', text: 'Tool call interrupted by a crash; no result was recorded.' }],
         isError: true,
-        error: {
-          message: 'Tool call interrupted by a crash; no result was recorded.',
-          info: { name: 'InterruptedError', code: 'interrupted' },
-        },
+        error: { name: 'InterruptedError', code: 'interrupted' },
       },
       surfaceOp: 'append',
       ...callSeq !== undefined ? { sourceEventSeqs: [callSeq] } : {},

packages/core/session/src/types.ts

@@ -243,12 +243,13 @@ export interface SessionEventMap {
    */
   'tool/call': { turn: number; step: number; callId: CallId; name: string; arguments: string }
   /**
-   * A completed tool call's model-facing result, canonical failure detail, and
-   * optional tool-private `meta` presentation payload. `meta` is opaque to the
-   * core (the producing tool owns its shape and reads it back in `presentResult`)
-   * but MUST be JSON-serializable: `Session.append` runtime-validates all event
-   * data with `isJsonValue`, so a non-serializable `meta` is rejected at the
-   * source, and the durable log reproduces the identical card on replay. Absent
+   * A completed tool call's model-facing result, optional internal failure
+   * identity, and optional tool-private `meta` presentation payload. `meta` is
+   * opaque to the core (the producing tool owns its shape and reads it back in
+   * `presentResult`) but MUST be JSON-serializable: `Session.append`
+   * runtime-validates all event data with `isJsonValue`, so a non-serializable
+   * `meta` is rejected at the source, and the durable log reproduces the
+   * identical card on replay. Absent
    * unless the tool attaches one (e.g. `dsh-tool-fs` carries its result-time
    * contextual diff here).
    */
@@ -258,7 +259,7 @@ export interface SessionEventMap {
     callId: CallId
     content: ContentBlock[]
     isError: boolean
-    error?: { message: string; info?: { name: string; code: string } }
+    error?: { name: string; code: string }
     meta?: JsonValue
   }
   /** Steering content injected between steps of a running turn. */

packages/core/session/tests/repair.spec.ts

@@ -64,7 +64,7 @@ describe('interruptedTurnClosers', () => {
     expect(closers.map(e => e.seq)).toEqual([3, 4, 5])
     const result = closers[0]!
     expect(result.type === 'tool/result' && result.data).toMatchObject({
-      turn: 2, step: 1, callId: CallId('call-1'), isError: true, error: { info: { code: 'interrupted' } },
+      turn: 2, step: 1, callId: CallId('call-1'), isError: true, error: { code: 'interrupted' },
     })
   })
 

packages/core/tools/src/index.ts

@@ -349,6 +349,25 @@ export class ToolOutputError extends HarnessError {
   }
 }
 
+/** Convert one projector exception into the canonical invalid-output failure. */
+function projectionError(toolName: string, projector: 'render' | 'presentationMeta', error: unknown): ToolOutputError {
+  return new ToolOutputError(toolName, [`output.${projector} failed: ${errorMessage(error)}`])
+}
+
+/** Snapshot one projector result before later durable-result materialization. */
+function snapshotProjection<T>(toolName: string, projector: 'render' | 'presentationMeta', candidate: T): T {
+  try {
+    const detached = snapshotJsonValue(candidate)
+    if (detached === undefined) {
+      throw new ToolOutputError(toolName, [`output.${projector} returned non-lossless JSON`])
+    }
+    return detached
+  } catch (error: unknown) {
+    if (error instanceof ToolOutputError) throw error
+    throw projectionError(toolName, projector, error)
+  }
+}
+
 /** Successful canonical tool execution, including its Native/model projection. */
 export interface ToolExecutionSuccess {
   readonly isError: false
@@ -1156,10 +1175,23 @@ export class ToolRegistry extends Service {
     const violations = validateJsonSchemaValue(tool.output.schema, detached, 'value')
     if (violations.length > 0) throw new ToolOutputError(tool.name, violations)
     const value = deepFreeze(detached as JsonValue)
-    const content = tool.output.render(exec.arguments, value)
-    const meta = exec.parent === undefined && tool.output.presentationMeta !== undefined
-      ? tool.output.presentationMeta(exec.arguments, value)
-      : undefined
+    let rendered: ContentBlock[]
+    try {
+      rendered = tool.output.render(exec.arguments, value)
+    } catch (error: unknown) {
+      throw projectionError(tool.name, 'render', error)
+    }
+    const content = snapshotProjection(tool.name, 'render', rendered)
+    let meta: JsonValue | undefined
+    if (exec.parent === undefined && tool.output.presentationMeta !== undefined) {
+      let projected: JsonValue
+      try {
+        projected = tool.output.presentationMeta(exec.arguments, value)
+      } catch (error: unknown) {
+        throw projectionError(tool.name, 'presentationMeta', error)
+      }
+      meta = snapshotProjection(tool.name, 'presentationMeta', projected)
+    }
     return this.markCanonical(this.materializeFinalResult({
       isError: false,
       value,

packages/core/tools/tests/tools.spec.ts

@@ -147,6 +147,7 @@ describe('ToolRegistry', () => {
     })
     expect(result.isError).toBe(true)
     expect(result.content[0]?.type === 'text' && result.content[0].text).toContain('Error:')
+    expect(result.error).toMatchObject({ info: { name: 'ToolOutputError', code: 'INVALID_TOOL_OUTPUT' } })
     expect(observedError).toBe(true)
   })
 
@@ -209,10 +210,10 @@ describe('ToolRegistry', () => {
     }))
 
     const result = await ctx.tools.execute({ callId: CallId(projector), name: `throwing-${projector}`, arguments: {} })
-    expect(result).toMatchObject({
-      isError: true,
-      error: { message: projector === 'render' ? 'renderer exploded' : 'metadata exploded' },
-    })
+    expect(result.isError).toBe(true)
+    expect(result.error?.message)
+      .toContain(projector === 'render' ? 'renderer exploded' : 'metadata exploded')
+    expect(result.error?.info).toEqual({ name: 'ToolOutputError', code: 'INVALID_TOOL_OUTPUT' })
     expect('value' in result).toBe(false)
   })
 

packages/fs/tool-fs/src/read-render.ts

@@ -138,6 +138,7 @@ export async function buildWindow(
       appendToLineBuffer(chunk.slice(startPos, newlinePos))
       flushLine()
       startPos = newlinePos + 1
+      if (acc.done) return finish(acc, request, displayPath)
     }
     appendToLineBuffer(chunk.slice(startPos))
   }

packages/mcp/mcp-client/package.json

@@ -28,14 +28,14 @@
   },
   "dependencies": {
     "@modelcontextprotocol/sdk": "^1.12.0",
-    "schemastery": "^3.18.0"
+    "schemastery": "^3.18.0",
+    "zod": "^4.4.3"
   },
   "devDependencies": {
     "@deepseek-ai/dsh-tools": "workspace:^",
     "@deepseek-ai/dsh-llm": "workspace:^",
     "@modelcontextprotocol/server-everything": "^2026.7.4",
     "@modelcontextprotocol/server-filesystem": "^2026.7.4",
-    "cordis": "^4.0.0-rc.7",
-    "zod": "^4.4.3"
+    "cordis": "^4.0.0-rc.7"
   }
 }

packages/mcp/mcp-client/src/tools.ts

@@ -14,6 +14,8 @@
 
 import { createHash } from 'node:crypto'
 import type { Client } from '@modelcontextprotocol/sdk/client/index.js'
+import { ListToolsResultSchema } from '@modelcontextprotocol/sdk/types.js'
+import { z } from 'zod'
 import type { Context } from 'cordis'
 import type { ToolDefinition, ToolExecution } from '@deepseek-ai/dsh-tools'
 import { assertSupportedJsonSchema } from '@deepseek-ai/dsh-tools'
@@ -46,6 +48,35 @@ const INVALID_NAME_CHARS = /[^A-Za-z0-9_-]/g
 /** Hex chars of the SHA-256 identity hash appended on lossy normalization. */
 const HASH_LENGTH = 12
 
+/** Raw result record: the bridge owns JSON-value validation after transport. */
+const RawCallToolResultSchema = z.record(z.string(), z.unknown())
+
+/** List without mutating the SDK's per-page output-validator cache. */
+function listToolsUncached(client: Client, cursor?: string) {
+  return client.request(
+    { method: 'tools/list', ...cursor === undefined ? {} : { params: { cursor } } },
+    ListToolsResultSchema,
+  )
+}
+
+/** Call without the SDK pre-validating an output schema the bridge may not support. */
+function callToolUncached(
+  client: Client,
+  rawName: string,
+  args: Record<string, unknown>,
+  exec: ToolExecution,
+  opts: ToolBridgeOptions,
+) {
+  return client.request(
+    { method: 'tools/call', params: { name: rawName, arguments: args } },
+    RawCallToolResultSchema,
+    {
+      ...exec.signal ? { signal: exec.signal } : {},
+      timeout: opts.toolCallTimeoutMs,
+    },
+  )
+}
+
 /**
  * Derive the model-facing public name for one MCP tool.
  *
@@ -73,7 +104,7 @@ export function publicToolName(serverName: string, rawName: string): string {
  *
  * Two phases keep the swap safe:
  *
- * 1. Fetch: drain `client.listTools()` pagination and build the full next
+ * 1. Fetch: drain uncached `tools/list` pagination and build the full next
  *    generation of `ToolDefinition`s under public names. Any failure here
  *    (network error, duplicate raw name in the server's list) rejects and
  *    leaves the previous generation registered untouched.
@@ -101,7 +132,7 @@ export async function syncTools(
   const definitions = new Map<string, ToolDefinition>()
   let cursor: string | undefined
   do {
-    const response = await client.listTools(cursor ? { cursor } : undefined)
+    const response = await listToolsUncached(client, cursor)
     for (const tool of response.tools) {
       const publicName = publicToolName(opts.serverName, tool.name)
       if (definitions.has(publicName)) {
@@ -114,7 +145,7 @@ export async function syncTools(
         description: tool.description ?? '',
         parameters: tool.inputSchema,
         output: createOutput(tool.name, supportedOutputSchema(tool.outputSchema)),
-        execute: createExecutor(client, tool.name, opts),
+        execute: createExecutor(client, tool.name, tool.execution?.taskSupport === 'required', opts),
       })
     }
     cursor = response.nextCursor
@@ -170,7 +201,7 @@ function createOutput(rawName: string, structuredSchema: JsonSchemaNode | undefi
         content: { type: 'array', items: {} },
         structuredContent: structuredSchema ?? {},
       },
-      required: ['content'],
+      required: structuredSchema === undefined ? ['content'] : ['content', 'structuredContent'],
       additionalProperties: false,
     },
     render(_args, value) {
@@ -182,9 +213,10 @@ function createOutput(rawName: string, structuredSchema: JsonSchemaNode | undefi
 
 /**
  * Create an execute function for one MCP tool. The executor closes over the
- * raw MCP tool name and calls `client.callTool` with it (never the public
- * name), with abort signal and timeout, then maps the result to harness
- * ContentBlocks.
+ * raw MCP tool name and sends an uncached `tools/call` request with it (never
+ * the public name), with abort signal and timeout, then maps the result to
+ * harness ContentBlocks. Owning the raw request prevents the SDK's internal
+ * per-page schema cache from pre-validating a different contract.
  *
  * When the MCP server returns `isError: true`, the executor throws so that
  * the ToolRegistry's catch path produces an `isError` result for the model.
@@ -192,33 +224,30 @@ function createOutput(rawName: string, structuredSchema: JsonSchemaNode | undefi
 function createExecutor(
   client: Client,
   rawName: string,
+  taskRequired: boolean,
   opts: ToolBridgeOptions,
 ): ToolDefinition['execute'] {
   return async (args: unknown, exec: ToolExecution) => {
+    if (taskRequired) {
+      throw new Error(`Tool "${rawName}" requires task-based execution, which this bridge does not support`)
+    }
     // The agent loop passes `JSON.parse(model_arguments)` which is usually an
     // object, but can be any JSON value if the model misbehaves (outputs a bare
     // string/number/null). Fallback to {} lets the MCP server produce a
     // specific "missing required param" error the model can learn from.
     const argsObj = (typeof args === 'object' && args !== null ? args : {}) as Record<string, unknown>
-    const result = await client.callTool(
-      { name: rawName, arguments: argsObj },
-      undefined,
-      {
-        ...exec.signal ? { signal: exec.signal } : {},
-        timeout: opts.toolCallTimeoutMs,
-      },
-    )
+    const result = await callToolUncached(client, rawName, argsObj, exec, opts)
 
     // The SDK may return a legacy `toolResult` shape; normalize to content array.
-    if (!('content' in result) || !Array.isArray(result.content)) {
+    if (!Array.isArray(result.content)) {
       const rendered: unknown = 'toolResult' in result
         ? JSON.stringify(result.toolResult)
         : '(no output)'
       const text = typeof rendered === 'string' ? rendered : '(no output)'
-      if ('isError' in result && result.isError === true) throw new Error(text)
+      if (result.isError === true) throw new Error(text)
       return {
         content: [{ type: 'text', text }],
-        ...'structuredContent' in result && result.structuredContent !== undefined
+        ...result.structuredContent !== undefined
           ? { structuredContent: result.structuredContent as JsonValue }
           : {},
       }
@@ -232,13 +261,13 @@ function createExecutor(
     const text = extractText(content, rawName)
 
     // MCP isError → throw so ToolRegistry produces an isError result for the model.
-    if ('isError' in result && result.isError === true) {
+    if (result.isError === true) {
       throw new Error(text)
     }
 
     return {
       content,
-      ...'structuredContent' in result && result.structuredContent !== undefined
+      ...result.structuredContent !== undefined
         ? { structuredContent: result.structuredContent as JsonValue }
         : {},
     }

packages/mcp/mcp-client/tests/mcp-client.spec.ts

@@ -1,4 +1,6 @@
 import { describe, expect, it, vi, beforeEach } from 'vitest'
+import { Client } from '@modelcontextprotocol/sdk/client/index.js'
+import { InMemoryTransport } from '@modelcontextprotocol/sdk/inMemory.js'
 import { Context } from 'cordis'
 import { CallId } from '@deepseek-ai/dsh-llm'
 import SystemPrompt from '@deepseek-ai/dsh-system-prompt'
@@ -14,6 +16,7 @@ interface MockTool {
   description?: string
   inputSchema: Record<string, unknown>
   outputSchema?: Record<string, unknown>
+  execution?: { taskSupport?: 'optional' | 'required' | 'forbidden' }
 }
 
 interface MockCallResult {
@@ -23,9 +26,26 @@ interface MockCallResult {
 }
 
 function createMockClient(tools: MockTool[], callResult: MockCallResult = { content: [{ type: 'text', text: 'ok' }] }) {
+  const listTools = vi.fn(async (
+    _params?: Record<string, unknown>,
+  ): Promise<{ tools: MockTool[]; nextCursor: string | undefined }> => ({ tools, nextCursor: undefined }))
+  const callTool = vi.fn(async (
+    _params?: Record<string, unknown>,
+    _compatibilitySchema?: unknown,
+    _options?: unknown,
+  ): Promise<Record<string, unknown>> => ({ ...callResult }))
   return {
-    listTools: vi.fn().mockResolvedValue({ tools, nextCursor: undefined }),
-    callTool: vi.fn().mockResolvedValue(callResult),
+    listTools,
+    callTool,
+    request: vi.fn(async (
+      request: { method: string; params?: Record<string, unknown> },
+      _schema: unknown,
+      options?: unknown,
+    ): Promise<unknown> => {
+      if (request.method === 'tools/list') return listTools(request.params)
+      if (request.method === 'tools/call') return callTool(request.params, undefined, options)
+      throw new Error(`unexpected MCP request: ${request.method}`)
+    }),
     setNotificationHandler: vi.fn(),
     connect: vi.fn().mockResolvedValue(undefined),
     close: vi.fn().mockResolvedValue(undefined),
@@ -205,6 +225,80 @@ describe('syncTools', () => {
     expect(ctx.tools.get('mcp__srv__page1')).toBeDefined()
     expect(ctx.tools.get('mcp__srv__page2')).toBeDefined()
   })
+
+  it('owns output validation independently of the SDK per-page cache', async () => {
+    const [clientTransport, serverTransport] = InMemoryTransport.createLinkedPair()
+    serverTransport.onmessage = (message) => {
+      if (!('id' in message) || !('method' in message)) return
+      const params = 'params' in message ? message.params : undefined
+      let result: Record<string, unknown>
+      if (message.method === 'initialize') {
+        const protocolVersion = params && 'protocolVersion' in params
+          ? params.protocolVersion
+          : '2025-11-25'
+        result = {
+          protocolVersion,
+          capabilities: { tools: {} },
+          serverInfo: { name: 'raw-test', version: '1' },
+        }
+      } else if (message.method === 'tools/list') {
+        const cursor = params && 'cursor' in params ? params.cursor : undefined
+        result = cursor === undefined
+          ? {
+            tools: [{
+              name: 'supported',
+              inputSchema: { type: 'object' },
+              outputSchema: {
+                type: 'object',
+                additionalProperties: false,
+                properties: { answer: { type: 'integer' } },
+                required: ['answer'],
+              },
+            }],
+            nextCursor: 'page-2',
+          }
+          : {
+            tools: [{
+              name: 'future-schema',
+              inputSchema: { type: 'object' },
+              outputSchema: { type: 'object', patternProperties: { '^x-': { type: 'string' } } },
+            }],
+          }
+      } else if (message.method === 'tools/call') {
+        const name = params && 'name' in params ? params.name : undefined
+        result = name === 'supported'
+          ? { content: [{ type: 'text', text: 'missing structured content' }] }
+          : { content: [42, null], structuredContent: ['kept', { nested: true }] }
+      } else {
+        result = {}
+      }
+      void serverTransport.send({ jsonrpc: '2.0', id: message.id, result })
+    }
+    await serverTransport.start()
+    const client = new Client({ name: 'cache-independent-test', version: '1' })
+    await client.connect(clientTransport)
+
+    try {
+      await syncTools(client, ctx, defaultOpts, new Map())
+
+      const missing = await ctx.tools.execute({
+        callId: CallId('missing'), name: 'mcp__srv__supported', arguments: {},
+      })
+      expect(missing.error).toMatchObject({ info: { code: 'INVALID_TOOL_OUTPUT' } })
+      expect(missing.error?.message).toContain('structuredContent')
+
+      const fallback = await ctx.tools.execute({
+        callId: CallId('fallback'), name: 'mcp__srv__future-schema', arguments: {},
+      })
+      if (fallback.isError) throw new Error('unsupported schema must use the bridge fallback')
+      expect(fallback.value).toEqual({
+        content: [42, null],
+        structuredContent: ['kept', { nested: true }],
+      })
+    } finally {
+      await client.close()
+    }
+  })
 })
 
 describe('tool execution', () => {
@@ -360,6 +454,21 @@ describe('tool execution', () => {
     expect('value' in result).toBe(false)
   })
 
+  it('rejects tools that require task-based execution', async () => {
+    const client = createMockClient([
+      { name: 'task-only', inputSchema: { type: 'object' }, execution: { taskSupport: 'required' } },
+    ])
+
+    await syncTools(client as never, ctx, defaultOpts, new Map())
+    const result = await ctx.tools.execute({
+      callId: CallId('task-only'), name: 'mcp__srv__task-only', arguments: {},
+    })
+
+    expect(result.isError).toBe(true)
+    expect(result.error?.message).toContain('requires task-based execution')
+    expect(client.callTool).not.toHaveBeenCalled()
+  })
+
   it('passes abort signal to callTool', async () => {
     const controller = new AbortController()
     const client = createMockClient(

packages/session-persistence/session-persistence/tests/contract.ts

@@ -149,7 +149,7 @@ export function runPersistenceContract(name: string, make: () => Promise<Contrac
         ])
         const synthetic = loaded.events.find(e => e.type === 'tool/result')
         expect(synthetic?.type === 'tool/result' && synthetic.data).toMatchObject({
-          callId: CallId('call-x'), isError: true, error: { info: { code: 'interrupted' } },
+          callId: CallId('call-x'), isError: true, error: { code: 'interrupted' },
         })
         // The synthetic result carries the SAME callId as the orphaned tool-call,
         // so deriveMessages() pairs them — no provider-invalid dangling call.

packages/support/invariants/src/index.ts

@@ -165,7 +165,7 @@ function validateEvent(trace: SessionTrace, event: SessionEvent): SessionTraceTr
       // A result needs a prior matching call in the same step. (The converse
       // does NOT hold: a call may have no result — a throwing tool-execution
       // pipeline step ends the turn with no tool/result, which is legal.)
-      const syntheticInterrupted = event.data.isError && event.data.error?.info?.code === 'interrupted'
+      const syntheticInterrupted = event.data.isError && event.data.error?.code === 'interrupted'
       if (!trace.pendingCalls.has(event.data.callId) && !syntheticInterrupted) {
         throw new InvariantError(`tool/result for ${event.data.callId} with no prior tool/call in this step`)
       }

packages/support/invariants/tests/invariants.spec.ts

@@ -217,7 +217,7 @@ describe('session-log invariants', () => {
         callId: CallId('crashed'),
         content: [{ type: 'text', text: 'interrupted' }],
         isError: true,
-        error: { message: 'interrupted', info: { name: 'InterruptedError', code: 'interrupted' } },
+        error: { name: 'InterruptedError', code: 'interrupted' },
       }, { surfaceOp: 'append' })
       session.append('step/end', { turn: 1, step: 1 })
       session.append('turn/end', { turn: 1, reason: { kind: 'interrupted' } })
@@ -500,7 +500,7 @@ describe('surface contract under the invariants composition', () => {
       callId: CallId('rewrite'),
       content: [{ type: 'text' as const, text: 'original' }],
       isError: true,
-      error: { message: 'exit 1', info: { name: 'ExitError', code: 'EXIT_1' } },
+      error: { name: 'ExitError', code: 'EXIT_1' },
       meta: { presentation: { kind: 'terminal', output: 'full output' } },
       futureField: { nested: ['preserve', 1] },
     }
@@ -585,7 +585,7 @@ describe('surface contract under the invariants composition', () => {
     ['callId', { callId: CallId('forged') }],
     ['turn', { turn: 2 }],
     ['step', { step: 2 }],
-    ['error', { error: { message: 'exit 1', info: { name: 'ExitError', code: 'DIFFERENT' } } }],
+    ['error', { error: { name: 'ExitError', code: 'DIFFERENT' } }],
     ['meta', { meta: { presentation: { kind: 'generic' } } }],
     ['future data', { futureField: { nested: ['changed'] } }],
   ])('rejects a content rewrite with altered %s', async (_label, altered) => {

pnpm-lock.yaml

@@ -1447,6 +1447,9 @@ importers:
       schemastery:
         specifier: ^3.18.0
         version: 3.18.0
+      zod:
+        specifier: ^4.4.3
+        version: 4.4.3
     devDependencies:
       '@deepseek-ai/dsh-llm':
         specifier: workspace:^
@@ -1463,9 +1466,6 @@ importers:
       cordis:
         specifier: ^4.0.0-rc.7
         version: 4.0.0-rc.7(@cordisjs/plugin-include@1.0.4)(@cordisjs/plugin-loader@1.0.0-rc.5)
-      zod:
-        specifier: ^4.4.3
-        version: 4.4.3
 
   packages/sandbox/sandbox:
     devDependencies: